Feeding speed online optimization method based on cutting force feed-forward-feedback control
By combining feedforward and feedback control in CNC machining, the equivalent cutting depth is estimated and the feed speed is adjusted, and the problem of insufficient cutting force stability in the prior art is solved, and the rapid and accurate optimization of feed speed and machining efficiency are achieved. promote.
Patent Information
- Application Number
- CN202411047909.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-11-15
AI Technical Summary
The existing online feed speed optimization method is insufficient in ensuring the stability of cutting force, resulting in limited processing efficiency and lack of rapid response optimization solutions.
Using a method based on cutting force feedforward-feedback control, the equivalent cutting depth is estimated through the feedforward controller and the feed speed is calculated. Combined with the classic PID controller to compensate for errors, the restraint adjusts the feed speed range to ensure that The peak cutting force is stable near the set value.
It realizes rapid and accurate optimization of feed speed, improves processing efficiency, and ensures the stability of the processing process and cutting force control accuracy.
Smart Images

Figure CN118963254A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a numerical control machining technology, in particular to an adaptive machining technology, specifically to an online feed speed optimization method based on cutting force feedforward-feedback control. Background Art
[0002] Machining efficiency is one of the important indicators to measure the performance of CNC machine tools. Since the actual machining process is a complex dynamic process with high nonlinearity and strong time-varying characteristics, in order to ensure the machining quality as much as possible, the process parameters are often determined conservatively offline, which limits the machining efficiency and does not give full play to the equipment potential. Among the process parameters related to the machining process, the feed speed is the adjustable process parameter that is most closely related to the machining efficiency. Increasing the feed speed can directly improve the machining efficiency of the machine tool, but the corresponding increase in cutting force is likely to cause cutting vibration, tool deformation and even tool damage, which will seriously affect the dimensional accuracy and surface quality of the parts. Therefore, by optimizing the feed speed to stably control the cutting force, the machining efficiency can be improved as much as possible while ensuring the machining quality.
[0003] The existing feed rate optimization methods for stable cutting force control can be divided into two methods: offline feed rate optimization and online feed rate optimization. Among them, the offline optimization method is limited by the accuracy of the cutting force theoretical model and the uncertain factors in the actual processing process, and the optimization results are often only of reference value. Among the online optimization methods, the paper "Adaptive execution of an NC program with feed rate optimization" collects cutting force data in real time during the actual processing process through a desktop dynamometer, and then optimizes the feed rate online through a fuzzy reasoning system to keep the cutting force stable. However, the fuzzy control method lacks clear interpretability and has poor accuracy and stability. The paper "An intelligent system for monitoring and optimization of ball-end milling process" monitors the cutting force online through a desktop dynamometer, and then optimizes the feed rate using a genetic algorithm. However, the iterative calculation amount of heuristic algorithms such as genetic algorithms is large, which reduces the real-time performance of feed rate optimization. Therefore, there is an urgent need for a method to achieve fast and accurate online optimization of feed rate with the goal of stabilizing cutting force. Summary of the invention
[0004] The purpose of the present invention is to address the shortcomings of the existing online feed speed optimization method with the goal of stabilizing cutting force, and to invent an online feed speed optimization method based on feedforward-feedback control of cutting force. The method takes the stabilization of the cutting force peak value within the optimization cycle near the set cutting force reference value as the control target, and adopts a method combining feedforward and feedback control to adaptively optimize the feed speed. The feedforward controller is composed of a series of data-driven proxy models based on the mapping relationship between cutting force and cutting parameters, which can directly estimate the current equivalent cutting depth, and then directly calculate a more reliable feedforward optimization value of the feed speed. The feedback control part adopts a classic PID controller, which is responsible for compensating the error of the feedforward optimization result to improve the control accuracy. Finally, the constraint constrains the feed speed value range and fluctuation range to generate the final command feed speed to ensure the stability of the machining process.
[0005] The technical solution of the present invention is: An online feed speed optimization method based on cutting force feedforward-feedback control is characterized in that: on the basis of monitoring the cutting force in the numerical control machining process, the peak value of the cutting force in the optimization cycle is stabilized near the set cutting force reference value, a feedforward controller composed of a cutting parameter estimator and a feed speed estimator is constructed in cascade, the feed speed feedforward instruction output by the feedforward controller and the feed speed feedback instruction output by the feedback controller are superimposed, and the final instruction feed speed is generated after the feed speed value range and the fluctuation range of the previous and next instructions are constrained by the instruction constraint; the two estimators in the feedforward controller are data-driven proxy models established using data generated by a cutting force theoretical prediction model, wherein the input of the cutting parameter estimator is the cutting force peak value and the real-time feed speed in the current optimization cycle obtained by monitoring, and the output is the estimated current equivalent cutting depth; the input of the feed speed estimator is the output of the cutting parameter estimator and the cutting force reference value, and the output is the estimated feed speed feedforward instruction; the input of the feedback controller is the difference between the cutting force reference value and the cutting force peak value in the current optimization cycle, and the output is the feed speed feedback instruction.
[0006] Among them, the cutting force refers to one of the cutting force in the feed axis direction, the cutting force resultant, the tangential cutting force or the spindle cutting torque; when the cutting force control object is selected as the cutting force in the feed axis direction, both estimators in the feedforward controller need to add the feed direction as input, and the added input is represented by the angle between the feed direction and the X-axis of the machine tool. The cutting parameter estimator and the feed speed estimator in the feedforward controller adopt, but are not limited to, data-driven agent models such as feedforward neural networks and radial basis function neural networks; according to the potential range of feed speed and cutting depth, the speed and cutting width are set to specified values close to the actual application process to generate a training data set, and the data set is used to train the corresponding cutting parameter estimator and feed speed estimator. The equivalent cutting depth refers to the equivalent cutting depth corresponding to the actual cutting force peak value and feed speed at the specified speed and cutting width. The feedback controller is optional and is suitable for scenes with high cutting force control accuracy requirements. Feedback controllers such as but not limited to PI controllers, PD controllers and PID controllers can be used.
[0007] It should be noted that the cutting force can also refer to the combination of the cutting force in the feed axis direction, the resultant force of the cutting force in the feed axis direction, and the tangential cutting force (spindle cutting torque). For this multi-cutting force target control scenario, it is necessary to build an independent feedforward controller and feedback controller for each control target based on the above method, and obtain independent command feed speeds through the command constraint, and then select the command feed speed that does not violate all other control targets and has the largest value as the final optimization result.
[0008] Among them, not violating all other control objectives means that the cutting force corresponding to the command feed rate corresponding to a certain cutting force control objective should not exceed the other cutting force control objectives; and a force estimator is further introduced for evaluation, and its construction method is the same as the two estimators in the feedforward controller; the input of the force estimator is the command feed rate to be evaluated and the estimated current equivalent cutting depth, and the output is the cutting force estimation value corresponding to the cutting force control objective to be evaluated.
[0009] In short, the feature of the present invention is to stabilize the cutting force peak value within the optimization cycle near the set cutting force reference value as the control target, and to adaptively optimize the feed speed by combining feedforward and feedback control. The feedforward controller is composed of a series of data-driven proxy models based on the mapping relationship between cutting force and cutting parameters, which can directly estimate the current equivalent cutting depth, and then directly calculate a more reliable feedforward optimization value of the feed speed. The feedback control part adopts the classic PID controller, which is responsible for compensating the error of the feedforward optimization result to improve the control accuracy. Finally, the constraint constrains the feed speed value range and fluctuation range to generate the final command feed speed to ensure the stability of the machining process.
[0010] The beneficial effects of the present invention are: Compared with the existing online optimization method of feed speed with the goal of stabilizing cutting force, the method disclosed in the present invention combines the feedforward control method based on the neural network proxy model and the cutting force feedback control method. The feedforward control part can provide a fast response of the optimal feed speed during the control process and directly calculate a more reliable feedforward optimization value of the feed speed. The feedback control part can compensate for the error of the feedforward optimization result to improve the final control accuracy. Ultimately, the feed speed with the goal of stabilizing cutting force can be optimized quickly and accurately, and the machining efficiency of the machine tool can be improved as much as possible while ensuring the machining quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Figure 1 Schematic diagram of a feed speed optimization method according to an embodiment of the present invention.
[0012] Figure 2 Schematic diagram of the variable cutting depth blank used in the experiment of the present invention.
[0013] Figure 3 This is the prediction effect of the cutting parameter estimator of the present invention.
[0014] Figure 4 This is the prediction effect of the feed speed estimator of the present invention.
[0015] Figure 5 Schematic diagram of actual feed speed and cutting force before and after feed speed optimization. DETAILED DESCRIPTION
[0016] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.
[0017] like Figure 1-5 shown.
[0018] This embodiment takes the cutting force in the feed axis direction as the control object, and adopts the following method: Figure 1 The feed rate optimization method shown in the figure. The experiment was carried out on a DMG DMU80P machining center, and the selected tool was a three-edge flat head end mill with a diameter of 20mm and a helix angle of 48°. The customized blank size is 135mm×135mm×50mm, and the material type is 7075-T7451 aluminum alloy. During the experiment, the feed rate optimization cycle was set to 0.5s, and the cutting force peak reference value was set to 500N. With the X-axis as the feed axis, the tool path length is 135mm, the cutting width is set to 5mm, the spindle speed is set to 1500rpm, and the unoptimized initial feed rate is 1700mm / min. The designed variable cutting depth blank is as follows Figure 2As shown, the actual cutting depth is 2mm, 3mm and 4mm at displacements of 0-45mm, 45mm-90mm and 90mm-135mm respectively. The stable control of cutting force when the cutting depth is uneven is achieved by optimizing the feed speed.
[0019] The experimental process first identified the cutting force coefficient of the experimental blank material, and established a cutting force theoretical model to generate a cutting parameter-cutting force peak data set. Then, a feedforward neural network was used to train the cutting parameter estimator and feed speed estimator proxy models. The prediction results are as follows: Figure 3 , Figure 4 As shown. It can be seen that the neural network-based proxy model can well establish the mapping relationship between cutting parameters and cutting force peaks. Based on the method disclosed in the present invention, a feed speed optimization system is built based on the neural network proxy model of the cutting parameter estimator and the feed speed estimator, and then the designed variable cutting depth blank is cut. The cutting force monitored by the desktop dynamometer is used as input to optimize the feed speed online, and finally the workpiece is cut again using the unoptimized feed speed under the same conditions. The cutting force control effect is verified by comparing the cutting forces in the two cutting processes, and the improvement effect of the processing efficiency is verified by comparing the processing time.
[0020] Verification effect Figure 5 As shown, it can be seen that before the feed speed is optimized, the cutting force peak value in the entire processing process has never reached the set reference value of 500N. After the feed speed is optimized, the cutting force peak value can be effectively controlled near 500N. Taking the processing time as the measurement indicator, the processing efficiency is improved by about 25%. This also means that when the process parameters set in the actual processing process are too conservative, this method can quickly increase the cutting force and stabilize it to near the reference value by increasing the feed speed, thereby shortening the processing time and improving the processing efficiency. Therefore, the present invention can achieve rapid and accurate optimization of the feed speed with the goal of stabilizing the cutting force.
[0021] The parts not involved in the present invention are the same as the prior art and are implemented by using the prior art.
Claims
1. A feed speed online optimization method based on cutting force feedforward-feedback control, characterized in that: On the basis of monitoring the cutting force in the CNC machining process, the control target is to stabilize the cutting force peak value within the optimization cycle near the set cutting force reference value. A feedforward controller composed of a cascade of a cutting parameter estimator and a feed speed estimator is constructed. After the feed speed feedforward instruction output by the feedforward controller and the feed speed feedback instruction output by the feedback controller are superimposed, the feed speed value range and the fluctuation range of the previous and next instructions are constrained by the instruction constraint to generate the final instruction feed speed; the two estimators in the feedforward controller are data-driven agent models established using data generated by the cutting force theory prediction model, in which the input of the cutting parameter estimator is the cutting force peak value and real-time feed speed obtained by monitoring within the current optimization cycle, and the output is the estimated current equivalent cutting depth; the input of the feed speed estimator is the output of the cutting parameter estimator and the cutting force reference value, and the output is the estimated feed speed feedforward instruction; the input of the feedback controller is the difference between the cutting force reference value and the cutting force peak value within the current optimization cycle, and the output is the feed speed feedback instruction.
2. The method for online optimization of feed speed based on cutting force feedforward-feedback control according to claim 1, characterized in that: The cutting force refers to one of the cutting force in the feed axis direction, the resultant cutting force, the tangential cutting force or the spindle cutting torque; when the cutting force control object is selected as the cutting force in the feed axis direction, both estimators in the feedforward controller need to add the feed direction as input, and the added input is represented by the angle between the feed direction and the X-axis of the machine tool.
3. The method for online optimization of feed speed based on cutting force feedforward-feedback control according to claim 1, characterized in that: The cutting parameter estimator and feed speed estimator in the feedforward controller adopt data-driven agent models such as but not limited to feedforward neural networks, radial basis function neural networks, etc.; a training data set is generated according to the potential variation range of feed speed and cutting depth and the rotation speed and cutting width are set to specified values close to the actual application process, and the data set is used to train the corresponding cutting parameter estimator and feed speed estimator.
4. The method for online optimization of feed speed based on cutting force feedforward-feedback control according to claim 1, characterized in that: The equivalent cutting depth refers to the equivalent cutting depth corresponding to the actual cutting force peak value and feed speed at a specified rotation speed and cutting width.
5. The method for online optimization of feed speed based on cutting force feedforward-feedback control according to claim 1, characterized in that: The feedback controller is not a mandatory option and is suitable for scenarios where high cutting force control accuracy is required. Feedback controllers such as but not limited to PI controller, PD controller and PID controller may be used.
6. The method for online optimization of feed speed based on cutting force feedforward-feedback control according to claim 1, characterized in that: The cutting force mentioned above refers to the cutting force in the feed axis direction, the resultant force of the cutting force in the feed axis direction and the combination of the tangential cutting force. For this multi-cutting force target control scenario, an independent feedforward controller and feedback controller should be constructed for each control target, and independent command feed speeds should be obtained after passing through the command constraints. Then, the command feed speed with the largest value that does not violate all other control targets is selected as the final optimization result.
7. The method for online optimization of feed speed based on cutting force feedforward-feedback control according to claim 6, characterized in that: Selecting a command feed rate that does not violate all other control targets and has the largest value means that the cutting force corresponding to the command feed rate corresponding to a certain cutting force control target should not exceed the reference values corresponding to the other cutting force control targets; A force estimator is further introduced for evaluation, and its construction method is the same as the two estimators in the feedforward controller; the input of the force estimator is the command feed speed to be evaluated and the estimated current equivalent cutting depth, and the output is the cutting force estimation value corresponding to the cutting force control target to be evaluated.
